The present disclosure applies to the field of biometric identification technologies and provides a method and an apparatus for fingerprint identification. The method includes: extracting a minutia of the input fingerprint image by using a statistical method; performing fingerprint matching according to the extracted minutia to obtain a fingerprint identification result. According to the method provides in embodiments of the present disclosure, the direction of the minutia is calculated by using statistical information, a descriptor with statistical significance is added for the minutia, and during the matching process, calculation of the similarity of the minutia by using the descriptor and region matching by using information of the direction field and the gradient field of the overlapping region are added, therefore, instability and weak specificity of expression of fingerprint characteristics in a conventional algorithm are avoided, and accuracy of the fingerprint identification is improved.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method for fingerprint identification, comprising: extracting a minutia of an input fingerprint image by using a statistical method; and performing fingerprint matching according to the extracted minutia to obtain a fingerprint identification result; wherein the extracting the minutia of the input fingerprint image by using the statistical method comprises: calculating a direction of the minutia by using regional statistical information; adding a descriptor with statistical significance for the minutia according to the direction of the minutia; and extracting a triangle vertex direction of the minutia; wherein calculating the direction of the minutia by using the regional statistical information comprises: selecting a statistical region of a certain minutia; calculating and weighting gradient directions of all minutiae within the statistical region; performing a statistics on a gradient direction distribution of all the minutiae within the statistical region according to the gradient directions and weighted values of all the minutiae within the statistical region; performing an interpolation operation according to the gradient direction distribution to obtain a direction distribution function; and finding a maximum of the direction distribution function to obtain a direction of the minutia, and returning to the step of selecting a statistical region of a certain minutia until all minutiae of the input fingerprint image are traversed.
2. The method of claim 1 , wherein the adding the descriptor with statistical significance for the minutiae comprises: selecting a descriptor extracting region of a certain minutia; determining a regional four-quadrant according to the descriptor extracting region and a direction of the minutia; calculating gradient directions of all minutiae of the regional four-quadrant and performing an interpolation operation; and performing a statistics on a gradient distribution of the regional four-quadrant after the interpolation operation, forming a descriptor of the minutia, and returning to the step of selecting a descriptor extracting region of a certain minutia until all minutiae of the input fingerprint image are traversed.
3. The method of claim 1 , wherein the performing the fingerprint matching according to the extracted minutia comprises: calculating a similarity of the minutia; when the similarity of the minutia meets a preset first threshold, adjusting the fingerprint image; calculating a similarity of an overlapping region; and when the similarity of the overlapping region meets a preset second threshold, determining that the matching is successful.
4. An apparatus for fingerprint identification, comprising a processor, wherein the processor is configured to extract a minutia of an input fingerprint image by using a statistical method; and perform fingerprint matching according to the extracted minutia to obtain a fingerprint identifying result; wherein the processor is configured to calculate a direction of the minutia by using regional statistical information; add a descriptor with statistical significance for the minutia according to the direction of the minutia; and extract a triangle vertex direction of the minutia; wherein the processor is configured to select a statistical region of a certain minutia; calculate and weight gradient directions of all minutiae within the statistical region; perform a statistics on a gradient direction distribution of all the minutiae within the statistical region according to the gradient directions and weighted values of all the minutiae within the statistical region; perform an interpolation operation according to the gradient direction distribution to obtain a direction distribution function; and find a maximum of the direction distribution function to obtain a direction of the minutia, and return to the step to select a statistical region of a certain minutia until all minutiae of the input fingerprint image are traversed.
5. The apparatus of claim 4 , wherein the processor is configured to select a descriptor extracting region of a certain minutia; determine a regional four-quadrant according to the descriptor extracting region and a direction of the minutia; calculate gradient directions of all minutiae of the regional four-quadrant and perform an interpolation operation; and perform a statistics on a gradient distribution of the regional four-quadrant after the interpolation operation, form a descriptor of the minutia, and return to the step to select a descriptor extracting region of a certain minutia until all minutiae of the input fingerprint image are traversed.
6. The apparatus of claim 4 , wherein the processor is configured to calculate a similarity of minutia; adjust the fingerprint image when the similarity of the minutia meets a preset first threshold; calculate a similarity of an overlapping region; and determine that the matching is successful when the similarity of the overlapping region meets a preset second threshold.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
July 6, 2016
November 21, 2017
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.